That's what I'd expect too - would love to see the evidence as well!

Also, many interesting related questions: is the quality of match & wages lower/higher with no relocation necessary? How much do firms take advantage of recruiting from anywhere?

I haven't written down a model, but I'd say if wage = marg. prod, we wouldn't expect this pattern

However, in the presence of search frictions, I wouldn't expect workers to be paid their marginal product anyway...

2/2

Great question! Could be several things:
- within occup, higher wages reflect better matches (wage = worker qual + firm qual + match qual)
- cost of firing (as you say)
- cost of worker relocation only pays back for high wage jobs

1/2

thank you Emma!

Finally, massive thanks to Kory Kroft, Davin Pope, Milena Djourelova, Ruben Durante and Gregory J. Martin for generously sharing their data on Craigslist expansion and posts!

... but at the same time, ~1/3 of jobs aren't online at all, with lower-wage, lower-educ occupations particularly underrepresented - making this question still relevant today.

If you want to learn more, including what Kijiji is, full paper available here: mariabalgova.github.io/JMP.pdf

TL;DR: I show online job boards made it easier for workers to find better opportunities across space, increasing wages and mobility as a result.

Ofc, hiring online today looks very different, with AI helping to pick best candidates and guide job search...

(iii) within cities, increase in migration inflows and increase in wages should be concentrated in the same occupations

I find that this is true for Craigslist cities but not for control cities, where the relationship is negative - driven by labour supply rather than by better matching.

Given these patterns in baseline recruitment, we'd expect

(i) Craigslist to have larger impact in occupations with high non-local recruitment in 1990

(ii) the impact to be larger for higher wages where the quality of match often matters most

This is exactly what I find!

The data shows that firms recruit non-locally to find specific talent rather than to hire cheaply:

- non-local vacancies are posted in large markets with higher wages & lower-unemployment

- firms target markets which specialise in the occupation they’re searching for

I address this 2nd challenge by exploiting the placement of help-wanted ads across US newspapers in 1990.

Because firms had to make conscious choice about which newspaper to advertise in, this data uniquely captures which jobs were posted non-locally and why.

Higher migration + higher wages suggest that Craigslist improved matching between cities.

However, without any data on firm recruitment, it's hard to identify what drives these results.

For all we know, the migration and wage effects might be caused by different mechanisms!

At the same time, average wages in the treated cities went up by about 0.8% annually, putting an end to convergence in pay.

I find that Craigslist entry into a city significantly increased geographic mobility in and out of the city.

Looking at changes in bilateral city-to-city migration flows, almost all of the increase in migration churn is driven by greater mobility between Craigslist cities.

First challenge: the internet radically changed many aspects of life at the same time, making it difficult to identify the effect of online recruitment alone.

My solution: focus on the staggered rollout of one of the first online job boards, Craigslist, across US cities.

We know the introduction of the internet improved efficiency in matching *within* local labour markets.

However, internet also kills distance --> should make it easier for firms and workers to match *between* markets.

My paper is the first to explore this channel empirically.